Amazon's trouble raises cloud computing doubts

SAN FRANCISCO: The black out at Amazon's EC2 (Elastic Cloud Computing) data centre has cast a shadow over cloud computing, which has been hailed as a sturdy, reliable and inexpensive storage and network solution, especially for small and medium enterprises (SMEs) that cannot afford their own large servers.

On the early morning of April 21 (Pacific Day Time), Amazon's EC2 data centre in Virginia crashed, taking down with it several popular websites and small businesses that depend on it. These included favoured social networking destinations like Evite, Quora, Reddit and Foursquare, among others. Now, the question is being asked: if an Amazonian cloud giant can crash so badly, what about the rest? Is cloud computing as reliable as we thought?

"People will now realise that cloud isn't magic like they earlier thought it was," says Lydia Leong, research vice-president and cloud computing expert at technology research and advisory firm Gartner. "They will now realise that cloud is merely about viability and not about continuous availability."

But that's exactly the kind of marketing pitch that sold cloud computing to many small businesses, including the ever-increasing social networking bandwagon. SMEs are now graduating to the next level of cloud computing, using it not just for storage, but also for active computing purposes like communication, sustaining remote workforces and deploying cloud services like remote IT help, cloud operating systems, and so on. The impact of such an outage, therefore, is felt even more.

Online businesses affected by the EC2 outage lost that many hours of ad revenues, business opportunities and drops of the precious trust of many loyal followers, a primary pillar of social networking. The losses are hard to quantify.

"Since Amazon isn't giving a customer list, we can only guess. From what we know, it's probably in millions of dollars," says computer scientist David Alan Grier of the eminent Institute of Electrical and Electronics Engineers ( IEEE) Society and author of When Computers were Human and Too Soon to Tell: Essays for the End of the Computer Revolution. "The biggest problem, though, could be the loss of confidence in cloud computing. We still don't know why the Virginia data centre failed."

The EC2 holds incredibly valuable data of Amazon's cloud client companies. And yet, Amazon's Virginia centre is, according to sources, remarkably open and vulnerable, located in an ordinary industrial building near Dulles Airport.

Although Amazon will probably recover quickly, the event has damaged its credibility. Time will tell how badly. "If Amazon can explain the problem and make a good case for why the damage may not be big, then it will be fine," says Grier. "If not, the work will go elsewhere. Amazon may be a big player, but there are other big players waiting to step into the game." These include the likes of Google, IBM, Cisco, RedHat and Microsoft (whose cloud ads are all over Silicon Valley), to name a few.

"Amazon's cloud competitors are likely to use this outage as a marketing tool. But it could have happened to anybody," says an ex-HP cloud veteran who currently works at one of Silicon Valley's most promising cloud start-ups, which recently got acquired.

There's an important lesson to be learnt for cloud users from this incident: diversification. "As a business, it makes sense to not depend on a single cloud provider or a single data centre alone," says the cloud veteran. "You would not put all your eggs in one basket, would you?" His company's major product launch almost got jeopardised due to the EC2 outage, but didn't suffer heavily because it is distributed over several cloud providers and data centres.

For the same reason, two start-ups of Nicolai Wadstrom, the serial entrepreneur and investor, didn't crash either. His social Internet start-up Urban Metrics as well as an Internet programmers' community deploy Amazon's EC2, but also their own dedicated servers. "After this crash, we will think more about redundancy," says Wadstrom. "That involves spreading servers on multiple availability zones and perhaps on different providers."

The message for cloud computing users, thus, could not have been clearer. They need to consider their own servers for certain operations and use multiple cloud providers and be able to seamlessly move between them. In true Silicon Valley style, this new need will probably give birth to another crop of start-ups. ( Witsbits, one of Wadstrom's new start-ups, already offers a technology to facilitate this seamless integration.)

The cloud veteran says the whole premise of cloud computing is throw-away cheap computing, storage and network resources. You just need to know how to use it," he says. "Most small businesses do not know how to build clustered or distributed solutions. So, in a way, Amazon can, and ought to, beef up its solutions so that the user is less affected by such an outage."